Bioresource Technology 98 (2007) 1529–1534

Removal of ammonia as struvite from anaerobic digester effluentsand recycling of magnesium and phosphate

Mustafa Turker a, Ipek Celen b,*

a Pak-Food Industries, P.O. Box 149, 41001 Izmit, Kocaeli, Turkeyb Department of Environmental Engineering, Gebze Institute of Technology, Gebze, Kocaeli, Turkey

Received 13 July 2004; received in revised form 1 September 2005; accepted 15 June 2006Available online 2 November 2006

Abstract

A second order kinetic model was developed to predict the rate and extent of NHþ4 removal as struvite from anaerobic digester efflu-ents. Alternative to this, NHþ4 can be recovered from struvite and the remaining Mg2+ and PO3�

4 can be recycled back to the wastewaterto fix more NHþ4 . The NHþ4 solution was retained and the remaining Mg2+ and PO3�

4 were returned back to be mixed with wastewater. Ina five-step process, NHþ4 recovery was initially 92% and progressively decreased to 77% in the fifth stage, due to loss of Mg2+ and PO3�

4 ateach step in the supernatant. Finally, economic analysis of recycling nutrients was performed and compared to the one step process. Thecost of NHþ4 recovery was calculated as $0.36/kg NH4-N which is lower than $7.7/kg NH4-N the cost of one step process without con-sidering the market value of struvite obtained in one step process.� 2006 Elsevier Ltd. All rights reserved.

Keywords: Struvite precipitation; Ammonia removal; Kinetics; Recyling; Economics

1. Introduction

Nitrogen compounds, especially NHþ4 , may be presentin domestic and some industrial wastewaters at levels thatmay deteriorate the quality of receiving waters. Increasingawareness of environmental issues in recent years hasresulted in a number of technologies being applied forthe removal of nitrogen from wastewaters. Some technolo-gies, such as nitrification/denitrification and breakpointchlorination, convert ammonia to di-nitrogen gas. How-ever an approach based on resource recovery is preferableand may contribute positively to an overall nitrogen bal-ance as well as to the economics of wastewater treatmentleading to sustainable technologies. In our previous workwe have studied the optimisation of ammonia recovery as

0960-8524/$ - see front matter � 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.biortech.2006.06.026

* Corresponding author. Present address: Biosystems Engineering andEnvironmental Science Department, The University of Tennessee, Knox-ville, USA. Tel.: +1 865 974 2152.

E-mail address: icelen@utk.edu (I. Celen).

struvite from anaerobic digester effluents according to fol-lowing reaction (Celen and Turker, 2001):

Mg2þ þNHþ4 þ PO3�4 þ 6H2O!MgNH4PO46H2O ð1Þ

The kinetics of homogeneous chemical reactions can bewritten with respect to one species as (Smith, 1970):

� d½C�dt¼ k½C�n ð2Þ

where C is the concentration of reactant, k is the rate con-stant, and n is the order of reaction. If the equation is inte-grated for the first, second, and third order, it yields thefollowing integrated equations respectively:

ln½C� ¼ ln½C�0 � kt for 1st order ð3Þ1

½C� ¼1

½C�0þ kt for 2nd order ð4Þ

1

2½C�2¼ 1

2½C�20þ kt for 3rd order ð5Þ

1530 M. Turker, I. Celen / Bioresource Technology 98 (2007) 1529–1534

When the left hand side of Eqs. (3)–(5) are plotted againsttime t, one gets straight line for correct reaction order.

2. Methods

2.1. Analytical methods

The wastewater used in this work to recover ammonia isthe effluent of anaerobic digester treating molasses basedindustrial wastewater, the chemical composition of whichis given in Table 1. All struvite precipitation experimentswere carried out in magnetically stirred (Ikamag, ModelREC – g, Germany) batch reactor equipped with tempera-ture control unit using 200 mL of the effluent. The reac-tions were performed in triplicate for raw wastewater andin duplicate for kinetic data. Also, the precipitation reac-tions were performed in triplicate for recycled and non-recycled nutrients. Since the anaerobic digester effluenttemperature was 37 �C, all precipitation experiments wereperformed at that temperature. A thermostatic controllerwas used to control the temperature of the reactor (Haakel,Model 001.3568 Germany) The pH measurements weremade with pH meter (Metrohm, Model E588, Switzerland)and pH was adjusted either with HCl or NaOH solutions.The chemicals used were commercial grade, and their con-tents were determined before the experiments. All analyseswere carried out according to Standarts Methods for theExamination of Water and Wastewater (1989). After eachexperiment, the supernatants and the precipitates were ana-lysed for Mg2+, NHþ4 and PO3�

4 to check the consistencyof experimental results. A mass balance confirmed thatthe Mg2+, NHþ4 , and PO3�

4 present in the waste wasaccounted for and exists in either in the supernatant orthe recovered precipitate.

2.2. Experiments for the kinetics of struvite formation

Kinetic studies related to struvite formation were per-formed with 200 ml wastewater at 37 �C. The experimentswere performed in duplicate and took place at room tem-perature. The wastewater was thoroughly mixed using amagnetic stir plate. pH was increased to 8.5 with NaOHand Mg2þ : NHþ4 : PO3�

4 molar ratio was adjusted to1:1:1. Initially phosphoric acid (H3PO4) was used as aPO3�

4 source and a predetermined amount of NaOH wereadded for pH adjustment. Since the reaction was very fast,

Table 1The approximate composition of raw wastewater

Parameter Concentration in water (mg l�1) and standard deviation

NHþ4 1400(±2.5)Mg2+ 21.4(±3)PO3�

4 24(±5)Ca2+ 21.2(±2)K+ 2150(±14)COD 3240(±15)

pH was not adjusted during the course of the reaction. Thereaction started immediately after adding MgCl2 Æ 6H2O asa Mg2+ source; and 5, 10, 20 and 30 s later samples weretaken and immediately filtered to determine the concentra-tion of Mg2+ in the supernatant. The Mg2+ concentrationwas used to follow the course of the reaction since itrequires a small amount (10 ml) of sample for analysis.

2.3. Experiments for the volatilization of ammonia from

precipitate

Three experiments were performed for the volatilizationof NH3 from precipitate; with and without the addition ofNaOH to the precipitate in a 110 �C oven and stripping ofNH3 from precipitate by distillation. In the first experi-ment, struvite was formed from the 200 ml wastewaterwhen Mg2þ : NHþ4 : PO3�

4 molar ratio was 1:1:1 at pH8.5. Ammonia was released into absorption water by add-ing NaOH and heating the solid struvite 3 h in a 110 �Coven. The molar ratio of NaOH:NHþ4 was 1:1 in order toneutralize NHþ4 -NH3. The second method for the vaporiza-tion of NH3 from precipitant was distillation method.While performing the experiments, NaOH was used toincrease pH because at high pH NH3 could be capturedin boric acid. In the last experiment, NaOH added precip-itate was boiled for 15 min. The molar ratio of NaOH:NHþ4was 1:1.

2.4. Experiments for the feasibility of ammonia recovery

from precipitated struvite

Mg2þ : NHþ4 : PO3�4 molar ratio and pH of the initial

reaction were 1.2:1:1.2 and 8.5, respectively. After the pre-cipitation experiment, NaOH was added to the struviteprecipitate (NaOH : NHþ4 ratio equal to 1:1), the mixturewas boiled and ammonia (NH3) was stripped to the atmo-sphere. The resulting material (recovered Mg2+ and PO3�

4 )was added to a subsequent batch reaction. This process wasrepeated four more times. After each experiment the super-natant was analysed for Mg2+, NHþ4 , and PO3�

4 . At theend, the precipitate was solubilized by adding concentratedHCl and analysed for Mg2+ and PO3�

4 in order to performeconomical analysis of the process.

3. Results and discussion

3.1. The composition of wastewater

The effluent contains approximately 1400 mg l�1 ammo-nium ðNHþ4 Þ, 2 mg l�1 phosphate ðPO3�

4 Þ and 21.4 mg l�1

magnesium (Mg2+). Since PO3�4 and Mg2+ concentrations

were in negligible amounts, laboratory experiments wereconducted using PO3�

4 and Mg2+ sources to force the pre-cipitation of NHþ4 as struvite and reduce the concentrationof NHþ4 in the anaerobic digester effluent. In an attempt tooptimize environmental conditions, Celen and Turker(2001) conducted experiments to identify the how much

y = -0.075x - 3.102

R2 = 0.8462

-6

-4

-2

0

0 10 20 30 40Time (second)

ln[M

g2+]

Fig. 1. Time versus ln[Mg2+] for 1st order reaction.

y = 4.9334x + 16.115

R2

= 0.9925

0

50

100

150

200

0 10 20 30 40Time (second)

1/[M

g2+]

Fig. 2. Time versus 1/[Mg2+] for 2nd order reaction.

y = 449.26x - 1198

R2 = 0.9298

-2000

3000

8000

13000

18000

0 10 20 30 40Time (second)

1/2

[Mg2+

]2

Fig. 3. Time versus 1/2[Mg2+]2 for 3rd order reaction.

Table 3Estimation of kinetic parameters in kinetic model

Order R2 k PredictedC0 (mg/l)

ExperimentalC0 (mg/l)

1st 0.8462 270 h�1 1090 17922nd 0.9925 17.76 · 103 l/mol h 1506 17923rd 0.9298 1.6 · 106 l2/mol2 h 486 1792

M. Turker, I. Celen / Bioresource Technology 98 (2007) 1529–1534 1531

PO3�4 and Mg2+ sources were added. Therefore, these ions

were added in equimolar quantities in struvite precipitationstudies due to equimolar presence of ions in the precipitate.

3.2. Kinetics of struvite formation

The time course of the reaction is given in Table 2 as theaverage of two sets of experimental data. The reaction wascompleted in about 30 s. Integrated forms of the first, sec-ond, and third order reaction models were fitted to theexperimental data. These are presented in Figs. 1–3. Theconstants of the kinetic model thus obtained are presentedin Table 3. The first order approach did not give satisfac-tory fit to the experimental data with the poor R-squarevalue of 0.84, whereas the third order gave reasonable fitwith the R-square value of 0.92. The best fit was obtainedwith the second order rate approach with the R-squarevalue of 0.99. The initial concentration of Mg2+ was betterpredicted by 2nd order model. The reaction of struvite for-mation was considered to be homogeneous and crystal for-mation was not taken into account. This assumption is truein the nucleation period of the reaction (see Fig. 4).

In the literature, there are a few reports dealing withstruvite kinetics. Gunn (1976) claimed that the order ofstruvite formation is 2.3. Ohlinger et al. (2000) reportedthat struvite crystallisation obeyed first order kinetics withrate constant 4.2 h�1. Nelson et al. (2003) also reportedfirst order kinetics with the rate constants 3.7 (at pH 8.4),7.9 (at pH 8.7) and 12.3 h�1 (at pH 9.0), in similar rangeas Ohlinger et al. (2000). However, we have observed muchfaster precipitation kinetics completed in less than a min-ute. The difference between our results and the othersmight be due to the fact that we have studied nucleationprocess whereas Ohlinger et al. (2000) and Nelson et al.(2003) have investigated crystal growth, both processesare governed by different mechanisms resulting in differentkinetic expressions.

3.3. Single step precipitation

In order to assess the economic viability of the processin comparison to existing nitrogen removal technologies,the cost of ammonia recovery as struvite was studied basedon the experimental results presented here. In this assess-ment, investment cost was not taken into account and only

Table 2Kinetic data (average of two sets of experimental data)

Time(s)

Mg2+ (mg/l)and standarddeviation

Mg2+

(M)RemainingMg2+ insupernatant(%)

ln[Mg2+]

1/[Mg2+]

1/2[Mg2+]2

0 1855.8(±12.9) 0.076 100 �2.6 13.1 85.45 558.9(±7) 0.023 30 �3.8 42.6 905.4

10 340.2(±10) 0.014 18.3 �4.3 71.4 255120 218.7(±12) 0.009 12.1 �4.7 107.5 578130 145.8(±3.5) 0.006 7.8 �5.12 166.7 13888.9

the cost of chemicals including NaOH and steam were con-sidered (Table 4). The commercial value of struvite wasalso not considered. However, the market prices of thechemicals used in the calculations are given in Table 4.

The results of the economic analysis for the experimen-tal results presented are given in Table 5, with the contribu-tion of each chemical to overall cost of struvite. The cost ofammonia removal is a function of the removal ratio andthe cost of chemicals added per kg ammonia present inthe effluent. The relative cost of MgCl2, as a Mg2+ source,is the highest compared to those of H3PO4 and NaOH,amounting to approximately 55–65% of the overall cost.

15 kg MgCl2.6H2O ($4.66)

Supernatant 6.15 L H3PO4 ($2.92)

16.95 kg MAP($0.62/kg MAP)($7.72/kg NH4

+-N)

18.25 L Caustic ($0.64)

Wastewater1 m3

NH4+ = 1366 mgl-1

Fig. 4. The simplified diagram for the calculation of cost of struvite fromthe experimental results.

Table 4The market prices of the chemicals used in the experiments

Chemical Price ($/kg)

H3PO4 (75%) 0.40MgCl2 Æ 6H2O 0.31MgO (85%) 0.44NaOH (100%) 0.12NHþ4 0.23

Table 5Economic analysis of the process

Product Price ($)/m3 wastewater

Consumed MgCl2 Æ 6H2O 0.00015Consumed 75% H3PO4 0.00021Consumed NaOH 0.00026Consumed energy 0.00012Recovered ammonium �0.00029

Total ($) 0.00045Total ($/kg NHþ4 ) 0.36 $/kg NHþ4

Table 6Economic analysis of subsequent precipitations using recycled and non-recycled nutrients

NH3 removal usingrecycled nutrients

NH3 removal usingnon-recycled nutrients

(NH3)initial (mg) 291 273.2(Mg)initial (mg) Negligible Negligible(PO4)initial (mg) Negligible Negligible(Mg)added (mg) 515 2187.5(PO4)added (mg) 1888 8555(Mg)final precipitate (mg) 458 2102.5(PO4)final precipitate (mg) 1367 7795Cost ($/kg NHþ4 ) 0.36 7.7

1532 M. Turker, I. Celen / Bioresource Technology 98 (2007) 1529–1534

However the cost of the ammonia recovery per kg ammo-nia fixed as struvite depends on the type and stoichiometricratio of chemicals used. When MgO is used as a Mg2+

source, the major cost is that of H3PO4 since MgO is rela-tively cheap compared to MgCl2. However, the cost of theprocess per kg ammonia removed as struvite does notchange much since the recovery of ammonia is relativelylow in comparison to the use of MgCl2. In order to reducethe overall cost of the process, cheaper caustic sources suchas Ca(OH)2 can be used for pH adjustment. In this case,calcium may compete for phosphate which may reducethe availability of phosphate for struvite formation(Stumm and Morgan, 1996).

The economic analysis based on the experimental resultspresented in the single step precipitation work compareswell to the cost reported by other researchers. The ammo-nium recovery cost was calculated to be $7.7/kg NHþ4 inthe one step process (Table 6) (Celen and Turker, 2001).Siegrist et al. (1994) considered electricity and maintenancein addition to chemical cost of the process and came upwith $9.10–11.38/kg NHþ4 -N. Andrade and Schuling(1999) reported the cost between $4.55–9.92/kg NHþ4 - Nat high ammonium concentrations. Siegrist (1996) calcu-

lated the cost of the process $9.72/kg NHþ4 -N. In all theseworks, including ours, the commercial value of struvite wasnot taken into account. Webb et al., 1995) reported the costof chemicals as $14.9/kg NH3-N. Struvite market is still notwell established. There is an active interest to produce stru-vite on commercial scale in Japan. Ueno and Fuji (2001)have reported the selling price of struvite approximately245 Euro/ton or $4.50/kg NHþ4 -N. As a result, local avail-ability and price of chemicals and use and selling price ofstruvite will determine whether this process can potentiallycompete with alternative nitrogen removal and recoverytechnologies currently available in the market.

Janus and van der Roest (1997) have compared econom-ics of currently available potential biological and physical/chemical treatment options for nitrogen-rich reject water.They are biofilm air-lift suspension reactor, conventionalnitrification and denitrification combined with membranebioreactor, bioreactor without sludge retention, ammoniastripping by both air and steam and magnesium phosphateprecipitation with optional recycle. They reported that low-est cost was obtained with bioreactor without sludge reten-tion between $3 and $5.5/kg NHþ4 -N. The cost of othermethods varied between $5 and $9/kg NHþ4 -N dependingon ammonia concentration present in the wastewater.

3.4. Multistep precipitation using recycled magnesium and

phosphate

The objectives of this work were NH3 removal fromwastewater, evaluation of subsequent precipitation reac-tions for NH3 removal using recycled nutrients (Mg2+

and PO3�4 ), and comparison of the above approach versus

NH3 removal using non-recycled nutrients. Some research-ers recycle magnesium and phosphate as a precipitationagent. Schulze-Rettmer et al. (2001) disintegrated precipi-tated struvite into ammonia, magnesium and phosphateby injection of hot steam and air. Then they stripped theammonia and condensed it in water. The residual magne-sium phosphate was used as a precipitating agent.

The vaporization of NH3 from precipitant with andwithout the addition of NaOH to the precipitant in a110 �C oven and vaporization (stripping) of NH3 from pre-cipitant by distillation is shown in Fig. 5. Ammonia

M. Turker, I. Celen / Bioresource Technology 98 (2007) 1529–1534 1533

decreases from the precipitant amended with NaOH at the110 �C oven, were 47.7%, 70%, and 81% after 1, 2, and 3 h,respectively. When the same experiment performed withoutaddition of NaOH to the precipitant at the 110 �C oven,ammonia decrease becomes less and takes more time. Asshown in Fig. 5, ammonia removal are 30%, 28%, 75%after 1, 2, and 24 h respectively. However, after 0.25 h,100% loss of ammonia from the precipitate was determinedwhen the distillation combined with NaOH wereperformed.

Since the wastewater contains high ammonium concen-tration (approximately 1400 mg l�1 NHþ4 ) the concentra-tions of phosphate and magnesium are very low,therefore they are added in equimolar quantities in struviteprecipitation studies. NH3 removal versus subsequent pre-cipitations using recycled and non-recycled nutrients isshown in Fig. 6. In a five-step process, NH3 recovery wasinitially 92% and progressively decreased to 77% in the fifthstage. NH3 removal decreased because a portion of theadded nutrients remained with the supernatant during eachsubsequent reaction; this reduced the molar ratio requiredfor complete NH3 removal. Also, recycled magnesiumphosphate becomes more inactive because Mg3(PO4)2 orMg4P2O7 contents were increased (Schulze-Rettmer et al.,2001). The final reaction using recycled nutrients removed20% less NH3 than reaction using non-recycled nutrients.

0

10

20

30

40

50

60

70

80

90

100

1 2 3 4 5Precipitation reaction

NH

3 re

mov

al (

%)

Using Recycled Nutrients Using Non-Recycled Nutrients

Fig. 6. NH3 removal versus subsequent precipitations using recycled andnon-recycled nutrients.

0

20

40

60

80

100

0.001 0.01 0.1 1 10 100

Time (hr)

NH

3 re

mov

al (

%)

NaOH amended precipitant at 110ºC

No NaOH amended precipitant at 110ºC

Distillation with NaOH amended precipitant

Fig. 5. The vaporization of NH3 from precipitant.

3.5. Economic analysis of recycling

Economic analysis of the process was carried out andcompared to the process using non-recycled nutrients. Inthis assessment, the cost of chemicals, Mg2+ source, PO3�

4

source and caustic and steam used for distillation havebeen considered in the calculations. As shown in Table 6,NH3 recovery cost from the wastewater is $0.36/kgNHþ4 -N. Ueno and Fuji (2001) has reported the marketvalue of struvite $4.50/kg NHþ4 -N. If this value is deducedfrom $7.7/kg NHþ4 -N which is the cost of single step pre-cipitation process, we obtain $3.2/kg NHþ4 -N as the netcost of single step struvite precipitation which is signifi-cantly higher than multi-step nutrient recycling process.As a result, ammonia recovery based on the recycling ofnutrients looks attractive and can be used to removeammonia especially from high strength nitrogenous waste-waters as a pretreatment step.

4. Conclusions

It can be concluded from the present investigation that:The rate of struvite precipitation is very fast and com-

pleted in less than a minute under the conditions studiedhere. The rate of reaction is closer to the 2nd order. Multi-step precipitation approach may provide a sustainablemethod to remove NH3 from wastewaters containing pri-marily high NH3 levels. This method may be suitable forhigh NHþ4 containing wastewaters for recovering bulk ofammonia. If the discharge standards are not met, remainingammonia may then be removed with conventional tech-niques in case single step struvite precipitation is not pre-ferred considering the uncertainties involved in the finaluse of struvite as fertilizer. Distillation combined with caus-tic addition is the efficient method for vaporization of NH3

from precipitation compared to those tested here. Usingrecycled nutrients for the precipitation process is as effectiveand less costly than using non-recycled nutrients.

Acknowledgements

The authors would like to thank Mr. Celal Aksit, PakFood Industry for laboratory analysis.

References

Andrade, A., Schuling, O., 1999. Literature search, digest and assessmentconcerning struvite recovery for recycling from waste waters, FinalReport for CEEP (CEFIC). Geochem Research BV/Utrecht Univer-sity Holland.

APHA, 1989. Standarts methods for the examination of water andwastewater, 17th ed. Amer. Public Health Assoc., Washington, D.C.

Celen, I., Turker, M., 2001. Recovery of ammonia from anaerobic digestereffluents. Environ. Technol. 22 (11), 1263–1272.

Gunn, D.J., 1976. Mechanism for the formation and growth of ionicprecipitates from aqueous solutions in far. Discuss. Chem. Soc., 133–140.

Janus, H.M., van der Roest, H.F., 1997. Do not reject the idea of treatingreject water. Water Sci. Technol. 35 (10), 27–34.

1534 M. Turker, I. Celen / Bioresource Technology 98 (2007) 1529–1534

Nelson, N.O., Mikkelsen, R.L., Hesterberg, D.L., 2003. Struvite precip-itations in anaerobic swine lagoon liquid: effect of pH and Mg:P ratioand determination of rate constant. Bioresour. Technol. 89, 229–236.

Ohlinger, K.N., Young, T.M., Schroeder, E.D., 2000. Post digestionstruvite precipitation using a fluidized bed reactor. J. Environ. Eng.126, 361–368.

Schulze-Rettmer, R., von Fircks, R., Simbach, B., 2001. MAP precipita-tion pilot plant investigation in Germany. In: Proc. 2nd InternationalConference on Phosphorous Recovery for Recycling from Sewage andanimal wastes. Noordwijkerhout, The Netharlands, March 12–14.

Siegrist, H., 1996. Nitrogen removal from digester supernatant-compar-ison of chemical and biological methods. Water Sci. Technol. 34, 399–406.

Siegrist, H., Gajcy, D., Sulzer, S., Roeleveld, P., Oschwald, R., Frischkn-echt, H., Pfund, D., Morgeli, B., Hungerbuhler, E., 1994. Nitrogen

elimination from digester supernatant with magnesium–ammoniumphosphate precipitation. In: Klute, R., Hahn, H. (Eds.), ChemicalWater and Wastewater Treatment II. Springer Verlag, Berlin, pp. 457–464.

Smith, J.M., 1970. Chemical Engineering Kinetics. McGraw Hill,Kogakusha Ltd.

Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry, third ed. WileyInterscience.

Ueno, Y., Fuji, M., 2001. Three years experience of operating and sellingrecovered struvite from full scale plant. Environ. Technol. 22, 1373–1381.

Webb, K.M., Bhargava, S.K., Priestley, A.J., Booker, N.A., Cooney, E.,1995. Struvite (MgNH4PO4 Æ 6H2O) precipitation: potential fornutrient removal and re-use from wastewaters. Chem. Australia 62,42–44.